Suggestion on Improving mRNA and sRNA mapping

Question

[kamel]

Hello friends I am new to the RNA-seq anaIyse have fastq files single end (1x50) of mRNA and sRNA (each file contains the reads of two organisms) ad and I want to study the expression of genes in each sample.

I made a mapping by STAR for the first sample train with the following command:

$ Linux_x86_64/STAR --runThreadN 12 --genomeDir  /index --sjdbOverhang 49 --readFilesIn /sample1.fastq.gz --readFilesCommand gunzip -c --outFileNamePrefix  /output -outSAMtype BAM SortedByCoordinate

but I got :

Uniquely mapped reads% | 58.83%

% of reads mapped to multiple loci | 38.77%

% of reads mapped to too many loci | 0.46%

I do not understand why I got this high percentage of reads mapped to multiple loci. Do you have an idea to improve the result of the mapping of mRNA in the STAR command that I used ???????

What is the best way to map sRNA reads to the reference genome ??

Thank you in advance

[h.mon]

Start troubleshooting your reads. Check for rRNA contamination, and check which genes have a particularly high number of multi-mappers.

[kamel]

can you give me more precision for the methodology I'm going to do

[h.mon]

Examine the alignments with IGV, or filter the multi-mappers with samtools (many methods for doing so, e.g. here, here, here and here) and examine the multi-mappers alignment with IGV.

Use bbduk with the ribokmers.fa.gz file to check for rRNA contamination.

[kamel]

Another question Plz. For sRNA mapping, I find somebody who selects reads from 18-30 and others from 18-26 before mapping. I want to know what size to select??

[grant.hovhannisyan]

You have mentioned that each file contains reads from two organisms. What are these organisms? If their genomes are similar, then you will get high number of mulitmaps.

[kamel]

these two organisms are not similar, it is a human genome infected by a pathogenic bacterium and therefore the mRNA has been sequenced it contains reads of bacteria and genome huamain (of course reads the human genome is more than bacterial genome ). I do not know is it normal to have the multimapped reads ?? or for a study of expression must I ask for sequencing paired end with a lognueur more than 50bp?

Answer 1

[kamel]

Another question Plz. For sRNA mapping, I find somebody who selects reads from 18-30 and others from 18-26 before mapping. I want to know what size to select??

[Asaf]

I don't think you need to select at all. Are your sRNAs from human or bacteria? If from bacteria then they should be long (up to ~150 nt)

Answer 2

[ataulhaleem]

sRNA less than 15bp dont make any sense. reads smaller than 15bp mostly might be part of RNA-degradom